Image Segmentation on PBD5K Difficult
99.13Pixel Accuracy (PA)MDCNeXt
Evaluation results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| MDCNeXtBackbone=ResNet-502025.08 | 99.13 | 62.52 | 62.45 | 66.11 | 77.26 | 97.58 | 74.59 | 65.22 | 14.61 | 0.86 | |
| MDCNetPub.=CVPR 2024, Backbone=ResNet-502025.08 | 99.08 | 56.52 | 54.2 | 53.72 | 70.84 | 95.38 | 67.09 | 56.9 | 17.61 | 0.94 | |
| SpiderPub.=ICML 2024, Backbone=ConvNeXt-L2025.08 | 99 | 44.12 | 47.4 | 44.12 | 60.21 | 93.14 | 60.23 | 49.21 | 23.44 | 1.29 | |
| SegFormerPub.=NeurIPS 2021, Backbone=MiT-B52025.08 | 98.88 | 33.9 | 35.12 | 32.1 | 39.02 | 89.21 | 57.21 | 36.7 | 33.03 | 1.53 | |
| ZoomNeXtPub.=TPAMI 2024, Backbone=ResNet-502025.08 | 98.88 | 50.12 | 50.79 | 49.36 | 67.85 | 95.08 | 62.89 | 51.82 | 21.61 | 1.17 | |
| DeepLabV3+Pub.=ECCV 2018, Backbone=ResNet-1012025.08 | 98.65 | 36.91 | 37.42 | 35.42 | 41.09 | 91.02 | 55.32 | 38.01 | 30.8 | 1.42 | |
| SAM 2Pub.=ICLR 2025, Backbone=Hiera-B+2025.08 | 98.36 | 19.89 | 17.82 | 16.78 | 50.29 | 73.02 | 57.8 | 10.11 | 44.74 | 1.87 | |
| SegGPTPub.=ICCV 2023, Backbone=ViT-L2025.08 | 98.17 | 14.99 | 13.59 | 13.37 | 52.81 | 68.67 | 53.35 | 13.47 | 42.33 | 1.82 |